Dementia Detection and Management Using Wearable Device fused Deep Learning Scheme

Abstract

Dementia affects over 50 million people worldwide, with numbers expected to
triple by 2050. Traditional diagnostic methods often lack early detection
capabilities and real-time monitoring, leading to delayed interventions and
increased caregiver burden. This study aimed to develop an integrated dementia
detection system combining wearable Internet of Things (IoT) with deep learning
for early identification and continuous monitor of dementia. The system has three
core components: (1) a wearable IoT using ESP32 and MAX30102 sensors to
collect data, (2) a deep learning scheme to compare five neural network approaches
(MLP, LSTM, GRU, CNN, and hybrid CNN-LSTM), and (3) a mobile app to ease
data visualization. The dataset comprised of 1,510 records with 11 features.
Preprocessing handled missing values, categorical encoding, and feature scaling;
while, SMOTE was used to address class imbalance. Results showed that the MLP
demonstrated superior performance with a 97% accuracy, 100% sensitivity, 94%
specificity, and 0.98 AUC-ROC. Device successfully collected physiological data as
displayed over mobile app – enabling real-time monitor and prediction. Device
yields a significant advancement in dementia care with early detection, continuous
monitoring, and improved accessibility. MLP offered exceptional performance
with the practical wearable implementation, provides a scalable solution for
healthcare systems seeking to improve dementia diagnosis and management while
reducing caregiver burden.